Documentation
Everything you need to put Focus, Turn and Pulse into a voice agent: integration guides, the wire protocol, the Python SDK and how we measure the models.
Start here Quickstart: license, install, first call in 10 minutes Pipecat, LiveKit, ElevenLabs bridge or raw WebSocket ›
Get started
Integrations
- Integration guidePut Tenms into Pipecat, LiveKit Agents, ElevenLabs Agents, the raw Edge WebSocket or the C ABI.
- Python SDKThe tenms package: in-process engine, licensing, telemetry, Pipecat, LiveKit and the Edge client.
- ElevenLabs bridgeRun ElevenLabs Agents through Tenms Edge: endpoint, session tokens, protocol mapping, pacing and gating.
API
Models & benchmarks
- Models and benchmarksHow Tenms Arena decides what ships, the Focus profile measurements, the open models and the training pipeline.
- Training reportTraining tenms-gru-v1 on a laptop: the model, the data, the run and what the benchmark shipped.
- Improving listening_miniWhy listening_pro still wins, the levers we will pull, the experiment plan and the gate every model must pass.
Security
Endpoints
| Service | URL | Serves |
|---|---|---|
| Edge | wss://edge.tenms.cloud | /v1/stream, /v1/elevenlabs/stream |
| Control plane and dashboard | https://app.tenms.cloud | session tokens, call summaries, stats |